Re: [boost] variant - off topic

Subject: Re: [boost] variant - off topic
From: Robert Ramey (ramey_at_[hidden])
Date: 2019-04-03 23:09:45

On 4/3/19 3:08 PM, Andrzej Krzemienski via Boost wrote:
> śr., 3 kwi 2019 o 23:28 Robert Ramey via Boost <boost_at_[hidden]>

>> So - simple proposal.  Require that sum types have operator= deleted.
>>
> 
> Interestingly, the current work on the language variant (
> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0095r2.html) goes
> into a somewhat similar direction. If types stored in the variant are
> unfriendly (strong guarantee cannot be offered) the assignment is
> implicitly declared as deleted. Users can still choose to define it and
> provide the implementation of their liking.

very interesting.  I checked the link.  Indeed, a cursory examination of 
the link convinces me that your abstract is likely correct.

Still, it still requires that the user of the library object investigate 
pretty deeply as to what the thing does behind the scenes - which is one 
thing I'm always criticizing.  One good thing about the current 
proposals is that they "just work (usually)" as there are no special 
rules to learn how to use them.

And of course whatever variant of variant I choose I can always just 
make it const so it can't be changed after initialization.

So maybe it just needs one of the &rzej; blog posts: title - "non const 
variant is code smell" and hope that it becomes a design pattern.

it also might be an interesting idea to create "safe_variant<T...>" 
which looks something like:

template<T ....>
struct safe_variant<T ...> : public std::variant<T ...> {
     operator=(const safe_variant<T ...> &) = delete;
}

Robert Ramey

> 
> Regards,
> &rzej;
> 
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
>